Title :
On modifications of Kohonen´s feature map algorithm for an efficient parallel implementation
Author :
Vassilas, Nilcolaos ; Thiran, Patrick ; Tenne, P.
Author_Institution :
NCSR Demokritos, Greece
Abstract :
Two new variants of Kohonen´s self-organizing feature maps based on batch processing are presented in this work. The purpose is to make available a finer grain of parallelism to be used in massively parallel systems. Ordering and convergence to asymptotic values for 1-D maps and 1-D continuous input and weight spaces are proved for both variants. Simulation on uniform 2-D data using 1-D and 2-D maps as well as simulations on 12-D speech data using 2-D maps are also presented to back the theoretical results
Keywords :
batch processing (computers); convergence; parallel algorithms; self-organising feature maps; speech recognition; 1-D continuous input; 1-D maps; 12-D speech data; 2-D maps; Kohonen´s feature map algorithm; asymptotic values; batch processing; efficient parallel implementation; massively parallel systems; uniform 2-D data; Computational modeling; Computer networks; Concurrent computing; Equations; Neural network hardware; Neural networks; Parallel processing; Process design; Speech; Throughput;
Conference_Titel :
Neural Networks, 1996., IEEE International Conference on
Conference_Location :
Washington, DC
Print_ISBN :
0-7803-3210-5
DOI :
10.1109/ICNN.1996.549022
